The study area covered the territory of the Republic of Ecuador (252,000 km²), located in the northwest of South America. Ecuador has four bioregions: The Coastal Region, the Andean Region, the Amazonian Region, and the Insular Region (i.e. Galápagos Islands). Generally, the ecosystems that characterize the Coastal region are the humid forests, dry forests, xerophytic scrub, and mangroves (Espinoza-Amén et al., 2021); the Andean, dominated by the Andes Mountains, is characterized by montane forests and paramos (Cruzatty & Vollmann, 2012; Varela & Ron, 2018); and the Amazonian by humid forests and lowland floodplain forests. The Galápagos Islands, situated ~1000 km from the Ecuadorian coast, have an arid climate along the coastline and semi-humid to humid as altitude increases (Varela & Ron, 2018).

The description of the species was based on seven collections of M. charantia deposited at the Herbarium of the Faculty of Natural Sciences of the University of Guayaquil (GUAY) and from at least ten living specimens observed in the field. This description helped us to avoid misidentification. The terminology used follows Jackson (1991). The habit is described, and also the organs of the vegetative and reproductive parts. Details under magnification have been observed in an Olympus SZ 61 0.67–4.5X stereoscope.

The occurrences of Momordica charantia in Ecuador were obtained from two online databases: Global Biodiversity Information Facility (GBIF) (https://www.gbif.org/country/EC/summary) and iNaturalist (https://www.inaturalist.org/). At the same time, an exhaustive search was carried out in several virtual herbaria, finding the scanned sheets not available in GBIF. Also, we included the information in the samples of M. charantia deposited at the Herbarium of the Faculty of Natural Sciences of the University of Guayaquil (GUAY). Due to the fact that GBIF has human observations from iNaturalist (photos taken from citizens), we used the RStudio (Racine, 2012) program to delete the duplicate occurrences (repeated information in GBIF and iNaturalist), thus getting a database with the following information: (1) occurrence ID, (2) geographic coordinates, (3) province, (4) locality, (5) year of collection or observation, and (6) occurrence type. All the occurrences were verified one by one, keeping only those with a clear image, for example, a photo of the living individuals or scanned herbaria sheets.

The information associated with the oldest record of M. charantia was used as an estimate of the year and place of its first introduction in Ecuador. This information allows to build the introduction and expansion history of the species through the different regions and provinces of Ecuador. Also, to analyze the spatial distribution and temporal change of the M. charantia records by bioregion in Ecuador, we made accumulative maps of the records organized by 45 years approximately. Also, the year reported for each record was used to elaborate a temporal accumulation curve. Additionally, to detect if M. charantia is present in protected areas, all the records of the species were overlapped on a layer of the National System of Protected Areas (available at http://ide.ambiente.gob.ec/mapainteractivo/), using QGIS software (v3.28.0).

To describe the climatic conditions associated with the M. charantia occurrences in Ecuador, we compiled for each record the values of six bioclimatic variables obtained from WorldClim v2.1 at the resolution of 2.5 minutes (Fick & Hijmans, 2017; available in http://www.worldclim.org/). These variables were: annual mean temperature (bio1), maximum temperature of warmest month (bio5), minimum temperature of coldest month (bio6), annual precipitation (bio12), precipitation of wettest month (bio13) and precipitation of driest month (bio14). Then a frequency histogram for each climatic variable was performed in RStudio v4.2.2 (Racine, 2012) considering the continental, insular (i.e. Galápagos), and the total records of M. charantia. To evaluate the differences in the climatic conditions where M. charantia is present among the regions of Ecuador (i.e. Andean, Coastal, Amazonian, and Insular), we ran a Principal Component Analysis (PCA) on 19 centered and scaled bioclimatic variables obtained from WorldClim v2.1 at the resolution of 2.5 minutes (Fick & Hijmans, 2017; available in http://www.worldclim.org/). Furthermore, to detect the type of natural ecosystems in which M. charantia occurs according to the records, we overlapped the layer of Ecosystems of Ecuador from the National Forest Monitoring System (MAE, 2013; http://ide.ambiente.gob.ec/mapainteractivo/), using QGIS software (v3.28.0). Also, we used the intersect tool of QGIS to obtain the number of occurrences in each ecosystem.

We performed a risk analysis (RA) for M. charantia using the protocol developed for Brazil and requested from the Instituto Hórus de Desenvolvimento e Conservação Ambiental – Risk Analysis adapted for Latin America (IHDCA, 2022; available at https://institutohorus.org.br/en/analise-de-risco-para-especies-exoticas-eng/), which is an adaptation of the Australian system (Pheloung et al., 1999; Pheloung, 2001). The main modifications in this RA were on the climatic questions and those focused on the oceanic islands (Ziller et al., 2018). As this RA excluded features specifically for islands, the protocol was applied only to continental Ecuador records. Also, considering that the climate questions were adapted for Brazil’s conditions, we adapted the climatic questions focusing on the bioregions of continental Ecuador. The RA protocol comprises three sections: (A) historical-biogeography, (B) undesired characteristics, and (C) biology and ecology characteristics, covering 45 questions, each one with a value assigned. The answer to each question must be supported by registering at least three references (scientific papers and technical reports). Not less than 30% of the questions per section should be answered to validate the RA. According to the score the level of risk is assigned with a recommendation from the system: (1) Very low risk (≤ 0), (2) Low risk (up to eight points), (3) Moderate risk (nine to 15 points), (4) High risk (16 to 30 points), and (5) Very high risk (>30 points). As recommendation, if the risk is very low or low, the species can be accepted; if the risk is moderate, the species might become invasive so more studies are required; if the risk is high or very high, the species should be rejected (Ziller et al., 2018). The institute mentions on its online page that the level of accuracy is approximately 90%. For Galápagos Islands, we did not find any available risk analysis protocol that could be applied to categorize M. charantia risk.

Trailing vine, to 6 m, branched; stem slender, costate, pubescent. Leaves alternate (Fig. 1A), blade 4–11 × 4–10 cm, suborbicular, 5(–7)–deeply lobed, foliaceous to chartaceous, pilose, deeply cordate, dentate, mucronate, veins pinnate, sulcate adaxially, prominent abaxially; petioles 1.5–5 cm (Fig. 1B). Tendrils simple. Flowers solitary (Figs. 1A, C), suberect; pedicels 4–10 cm; bracts reniform (Fig. 1C). Staminate flowers with hypanthium pilose, cylindric-campanulate; calyx lobes 2.5–4 mm, lanceolate; corolla 5–lobed, yellow, short-pilose; filaments 1–1.5 mm; anthers 2–2.5 mm. Pistillate flowers with ovary fusiform, tomentulose, tuberculate; style 2–3.5 mm; stigma 3, 2–lobed. Fruit fleshy capsules, fusiform, 3–5 cm, dehiscent in 3 recurved valves, pericarp muricate, rostrate, orange (Fig. 1D–F); seeds many, 8–10 × 4–6 mm, oblong, slightly compressed, aril sticky, bright-red (Fig. 1G).

A total of 311 verified records were obtained for M. charantia in Ecuador; 83% of the records (N = 259) were based on human observation and come from citizen science conducted through the iNaturalist platform (where 70% of these records are available only in iNaturalist due to their quality grade as “Needs ID”); and 17% (N = 52) are preserved specimens supported in herbaria (Appendix 1). Half of all records were in the Coastal Region (52%; N = 158), representing the highest number of occurrences per region, followed by the Insular Region (26%; N = 83). The Amazonian and the Andean regions had the lowest number of records (12%; N = 38 and 10%; N = 32, respectively). Although the Coastal Region had the greatest number of records, at a province level, Galápagos (Insular Region) had the greatest number of records per province (27%; N = 83), followed by Guayas and Manabí provinces (both from the Coastal Region) (22%; N = 69; 12%; N = 37 respectively) (Table 1).

According to the data obtained for Ecuador, the oldest record of M. charantia in Ecuador was in Guayaquil City (Ecuadorian Coast) in March 1846 (Fig. 2A). This first record corresponds to a preserved specimen at the Smithsonian National Museum of Natural History (available in http://n2t.net/ark:/65665/33571c7bc-53c6-462e-be5f-857650920ea4). The second oldest record was reported seven years later (1853) on Santa Cruz Island (Galápagos, Insular Region, Fig. 2A), which is available in the botanical collection database of the Swedish Museum of Natural History (S) (available in https://herbarium.nrm.se/specimens/S10-14834). Until 1923, records increased up to four (Fig. 2B). The third (available in http://n2t.net/ark:/65665/3d4410b96-038b-4209-b2dc-253fcf06889d) and fourth (available in http://n2t.net/ark:/65665/309cd28c6-7a61-4298-b4d7-ee1eb6495940) records were 70 years later (1923) on the Ecuadorian Coast again (Guayaquil and Milagro cities, respectively); both come from preserved specimens at the Smithsonian National Museum of Natural History. In 1977, ten records were added, making a total of 14 records (Fig. 2C). Two of these records were for the first time in the Amazonian Region (Napo), in 1966 and 1968. In less than 45 years, 297 records were added, with an occurrence for the first time in the Andean Region in 1990 (Santo Domingo de los Tsáchilas) (Fig. 2D). Figure 2E shows the temporal accumulation of M. charantia records in Ecuador from March 1846 to November 2022. Until the 20th century, there were no more than 51 records. In fact, since 2019, there has been an abrupt increase in the records (Fig. 2E).

Momordica charantia was detected in some priority areas of conservation in Ecuador. Data showed that 36% (N = 112) of the records of the species were distributed in five Biosphere Reserves (Bosque Seco, Chocó Andino de Pichincha, Galápagos, Podocarpus El Cóndor, Sumaco; Fig. 3A). Nine percent (N = 28) of the records were in nine Protected Areas (Cayambe Coca National Park, Galápagos National Park, Samanes National Recreation Area, Machalilla National Park, Mache Chindul Ecological Reserve, Manglares Churute Ecological Reserve, Manglar del Estuario del Río Muisne Wildlife Refuge, Parque El Lago National Recreation Area, Puerto Cabuyal-Punta San Clemente Marine Reserve; Fig. 3B). Three records were in Protective Forests (Daule-Pereira, Loma de Guayabillas, La Ceiba; Fig. 3C) and three in Ramsar Wetlands (Churute Wetland, Segua Wetland; Fig. 3D). The protected area with the highest number of records (N = 83) was the Galápagos Biosphere Reserve.

Momordica charantia records in Ecuador are frequently in areas with an annual mean temperature between 21°C to 26°C (Fig. 4A), this interval being larger for the continent (22–26°C; Fig. 4B) than for Galápagos (21–23°C; Fig. 4C). A similar pattern was observed for the maximum temperature of the warmest month and the minimum temperature of the coldest month (Fig. 4D–I). In the continent, the records of M. charantia were located in areas with a maximum temperature of the warmest month from 28–32°C and a minimum temperature of the coldest month from 14–20°C (Fig. 4E, H). While, in Galápagos, M. charantia records were in areas with maximum temperature of the warmest month from 28–31°C and minimum temperature of the coldest month from 15–19°C (Fig. 4F, I).

The records of this species in Ecuador are in areas with an interval of annual precipitation (196–4374 mm; Fig. 4J). In the continent, the species occupied more humid areas (196–4374 mm; Fig. 4K) compared to Galápagos (252–2676 mm; Fig. 4L). The same pattern occurred for the variables precipitation of the wettest month (Figs. 4M–O) and precipitation of the driest month (Fig. 4P–R). Values of bio13 and bio14 were significantly lower for Galápagos than the Continent (Fig. 4M–N; Q–R). For example, for the continent, the bio14 values were lower than 150 mm (Fig. 4O), while for Galápagos, the values of this variable were below 50 mm (Fig. 4R).

The first and second PCA axes (Fig. 5) explained 80% of the total variance. Positive values on principal component 1 (PC1, x-axis, 44.6% of variance) corresponded to areas with higher annual precipitation (bio12), while negative values corresponded to areas fluctuating in the seasonal and annual temperatures (bio4 and bio7, respectively; Fig. 5). Along principal component 2 (PC2, y-axis, 35.4% of variance), the positive values corresponded to areas with higher temperatures (e.g. bio1) (Fig. 5). Globally, the PCA showed that the records in Andean region occupy a broader range of climatic conditions, in contrast to the narrower climatic conditions of the records in Galápagos (Fig. 5). Momordica charantia occupied areas with higher rainfall in the continent, particularly in the Amazonian region, compared to Galápagos, where the species occupied drier climates. The areas inhabited by the species in Galápagos also had lower temperatures and greater annual and seasonal variation of this climatic variable (Fig. 5).

Momordica charantia was present in 15 types of natural ecosystems with 34 occurrences; 12 of the 15 ecosystems belonged to the Coastal Region (N = 31), two to the Andean Region (N = 2), and one to the Amazonian Region (N = 1) (Table 2). The ecosystems with the highest number of occurrences were the deciduous and semi-deciduous forests of the Jama-Zapotillo lowland, with seven records each and both with a high threat level (Table 2). The rest of the occurrences (N = 277, 89.1%) were in disturbed areas, being most of them urban or agricultural-industrial zones.

Momordica charantia was assessed as “High risk” (total score: 27) (Table 3). Forty of the 45 questions of RA were answered (Appendix 2). Enough information was available to answer the minimum number of questions per section. We answered 100% (N = 12) of the questions in the A section (biogeographical history), 92% (N = 12) of the questions in the B section (undesirable traits), and 80% (N = 16) of the questions in the C section (biological and ecological traits), with a total of 40 (89%) questions answered, with three references as minimum per question (Table 3; Appendix 3).

The A, B, and C sections contributed 52% (score: 14), 26% (score: 7), and 22% (score: 6) of the total score (total score: 27), respectively (Table 3). The A section obtained a score of 14 out of 17, representing 82% of the score. Climatic similarity and introduction history contributed significantly to the total score (Appendix 2). In section B, the score obtained was seven out of 14. The species had undesirable traits that contributed to this score, such as the possibility of being a weed in crops and toxic to animals or humans (Appendix 2). In section C, the score obtained was six out of 19. Seed dispersal by animals and short juvenile period were examples of reproductive traits that contributed to the score achieved in the C section (Appendix 2).